Issue with PromptFlow endpoint/deployment not working

Question

Hi Azure Community,

I'm currently working on a project PromptFlow, and I've encountered a persistent issue across different environments. Here’s a summary of the problem:

1. I cloned an existing, working PromptFlow and made modifications to the prompt. It initially ran successfully. However, after deployment, testing resulted in an error.
2. I made adjustments based on the error messages, but the issue persisted.
3. I then reverted to the original, previously working PromptFlow (without any changes), but it also returned the same error that I had encountered earlier.
4. To troubleshoot, I recreated the PromptFlow in Azure AI Studio, hoping the platform change would solve the issue, but the error occurred again.
5. In Azure Machine Learning Studio, I built a new PromptFlow from scratch, keeping the inputs and outputs minimal. This worked fine. However, when I expanded the flow slightly, the error resurfaced.
6. I further simplified the flow and adjusted the inputs and outputs, but the error remains unresolved.

Error -

At this point, I'm uncertain how to move forward. Has anyone experienced similar issues with PromptFlow in Azure or have suggestions for troubleshooting? Any input would be appreciated.

Answer 1

Hello Kira0308,

Welcome to the Microsoft Q&A and thank you for posting your questions here.

Regarding your explanation, the issue with your PromptFlow project could be resolve by the followings:

1. The compute session you're using needs to be updated with latest base image version. Even, if you are using customized base image, update the PromptFlow runtime to the latest version, rebuild the base image, and redeploy the flow https://learn.microsoft.com/en-us/azure/ai-studio/how-to/prompt-flow-troubleshoot
2. Also, update the tool names in your flow.dag.yaml file in raw code view https://learn.microsoft.com/en-us/azure/machine-learning/prompt-flow/troubleshoot-guidance?view=azureml-api-2

In addition to the links above, At 11:41PM on October 1, 2024. @Kira0308 confirmed it worked by:

I had to add disable trace = False in python script and it worked. Thanks!!

I hope this is helpful! Do not hesitate to let me know if you have any other questions.

---

Please don't forget to close up the thread here by upvoting and accept it as an answer if it is helpful.

Answer 2

I had to add disable trace = False in python script and it worked. Thanks !!

Answer 3

@Kira0308

Thanks for reporting this issue, product team has acknowledged this issue and working on a fix. Will update here again if the issue has any progress.

At this moment, please try the workaround to disable tracing and redeploy the flow by setting the env var "PF_DISABLE_TRACING" to true, this can avoid the potential error at this time.

os.environ['PF_DISABLE_TRACING'] = 'true'

I hope this helps!

Regards,

Yutong

-Please kindly accept the answer if you feel helpful to support the community, thanks a lot